Thelma & Louise ends with two women driving off a cliff holding hands. The car leaves the rocky edge. The camera stays with it. We see the car suspended in mid-air. We don't see it hit the ground. The film cuts before the impact. Matt Dicks: The interesting thing about that movie is it ends with the car mid-flight. We don't actually see it hit the ground, which allows the audience to think maybe they survived. It's improbable and it's not meant to be thought that way — but there's a reason why we don't see the car disappear off screen, and it's because we're going to hang it right here. We're going to stop the movie right here and you get to determine what you feel about it.1
Butch Cassidy and the Sundance Kid does the same thing. The two outlaws run out of the rocks to face a hundred federales. We know they're going to die. The film freezes the frame as they run out. We never see the bullets connect. So in our hearts and our minds they're still alive. Maybe Butch and Sundance got out of it.2
That's the frayed ending. Dicks: I think of it as a rope. I don't want a knot at the end of the rope. I want a frayed ending that allows people to sort of wonder what the hell happened the next day. That's great when they're wondering the next day. I'm thrilled.3 The knot is the closed ending — everything resolved, nothing left to wonder about, the audience can put the story away and stop thinking. The frayed ending leaves loose threads — small unanswered questions that keep the audience mentally returning to the piece after it's over.
The trick is not withholding the ending. Both films deliver their endings clearly (the women drive off; the outlaws face the federales). The trick is withholding the closure. The audience knows what happened. They don't quite know how to put it down.
A frayed ending has three properties:
1. The ending is delivered. The story reaches its destination. The 5-second moment of change has landed. The audience experiences the inversion. This is not a frayed story — the story itself is complete.
2. Specific small questions are left open. Not the central question (that's been answered) — peripheral questions. Did they survive somehow? What happened next morning? What did the other person think? These questions are small enough that the story works without their answers, but specific enough that the audience can't help wondering.
3. The audience continues to think about the piece after it's over. This is the operational target. That's great when they're wondering the next day. I'm thrilled.4 A piece that closes completely is finished the moment delivery ends. A frayed-ending piece keeps running in the audience's mind for hours, sometimes days. The continued thinking is what makes the piece stick in a way knotted-ending pieces don't.
The principle is additive to story-structure, not a substitute for it. The contrast arc still has to traverse, the 5-second moment still has to land, the inversion still has to feel inevitable. Then — after all that — the storyteller chooses whether to close completely (knot) or leave loose threads (fray). The frayed-ending choice is made at the end of crafting, not at the beginning.
The mechanism is cognitive non-completion. The brain remembers incomplete patterns longer than completed ones. Closed loops get filed away as resolved; open loops stay in working memory because the brain keeps trying to close them. The Zeigarnik effect names this in psychological literature; the frayed-ending principle exploits it in storytelling.
When Dicks chooses to end the spoon-of-power story with the pandemic-year compression rather than a clean explanation of where Jamie is now and what the spoon has done since 2007, the choice is deliberate. He could close the loop completely (Jamie's now 25 and working in finance, and the spoon sits in my classroom drawer between deployments). Closing it that way would resolve the audience's curiosity. They would put the piece away. The frayed version leaves Jamie's adult fate slightly unspecified — what is he like as an adult, does he remember the spoon, what does his version of the story sound like? The audience can't help thinking about these questions even though the central spoon-narrative has already landed.
The frayed ending also produces retelling. Audiences who continue thinking about a piece are more likely to tell other people about it. The thinking-process generates new angles, new memories of the piece, new things to say about it. Word-of-mouth spread of stories is correlated with how often audiences mentally return to them. Frayed endings produce return-thinking; knotted endings don't.
This is why Dicks frames the principle as thrill rather than as technique: I'm thrilled when they're wondering the next day. The continued audience thinking is the goal of the entire craft — every other Dicks principle is in service of producing pieces that audiences carry forward. The frayed ending is the last engineered move that maximizes this carrying-forward.
Frayed endings work because the 5-second moment of change has already landed. Without the 5-second moment, a "frayed ending" is just an unresolved ending — the audience experiences it as the storyteller having failed to finish. The 5-second moment is what gives the storyteller permission to fray.
The principle complements surprise in the Mount Rushmore: surprises that feel inevitable land with a thud; surprises that leave a small loose thread keep echoing. The combination — surprise plus small fray — produces the most durable audience-recall.
Frayed endings pair with end with heart not joke: heart-endings tend to fray naturally (emotional realizations are rarely fully closeable), while joke-endings tend to knot (the punchline closes the loop). The two principles reinforce each other.
The technique also complements small moments are more powerful: small moments naturally produce frayed endings because their resolutions are interior rather than external. Dramatic resolutions tend to knot (the car crash either kills you or doesn't); interior resolutions tend to fray (the realization is real but its consequences keep unfolding).
Dicks's spoon-of-power story has an explicit frayed ending. The story closes with the pandemic-year material:
The 2020-2021 school year was the hardest I've ever taught in my 26 years of teaching. The pandemic. We went right back to school in September and masks and social distancing and everyone was afraid... and I used that spoon more often that year than I've ever used it in my life. Every day people were wearing that spoon, and for the first time in my life my colleagues were wearing the spoon of power to get through the day. And as hard as it was, there was the best year of teaching I've ever had... I've always felt like I was the luckiest teacher in America because I have that spoon. I've had it for 16 years. It's literally in that bag right there. I carry with it. I carry it with me everywhere I go.5
What's open at the end:
Each of these is a small unanswered question. None of them prevents the story from landing. The central 5-second moment (a kitchen spoon became the most powerful teaching tool I'll ever own) has fully landed. But the peripheral questions stay open and the audience continues thinking about the piece after it ends.
The contrast with a knotted version: Dicks could have closed by explicitly explaining the spoon's full history since 2007, naming the colleagues who benefitted in 2020-21, and announcing where Jamie is now (presumably a college graduate, possibly with kids of his own). This version would close every loop. The audience would file the piece away as complete. The frayed version refuses these closures and keeps the audience returning.
You have a draft. The 5-second moment lands. The contrast arc completes. Now you're going to audit the ending for fray.
Read your final paragraph aloud. Now write down every question your ending fully answers: what happened next, what people did afterward, where everyone ended up, what it all meant in the long run. Make the list as complete as you can.
Look at the list. For each fully-answered question, ask: does the story need this answer for the 5-second moment to land? For most of them, the answer is no. The 5-second moment landed three paragraphs ago. The closure-content is decoration.
Now selectively remove closure-content. Not all of it — you want the audience to feel the story is complete, not unfinished. But specific decorative-closures can be cut. The version of Thelma & Louise that lands harder is the one where the car never hits the ground. The version of Butch Cassidy that travels is the one with the freeze-frame. Cut similarly. Leave the small loose threads.
The criterion: each remaining loose thread should be a question the audience might think about the next day but isn't actively required to resolve. Don't fray on the central plot question (the audience needs that closed). Fray on peripheral questions (those keep working in audience memory after the piece is over).
This audit takes 15 minutes per piece. Most amateur drafts over-close — they explain too much at the end. The fray audit teaches the storyteller to recognize over-closure and selectively remove it. Working storytellers report this as one of the highest-ROI revision techniques because the cost is small (trim some sentences) and the benefit is large (audience thinks about the piece longer).
You fray the central plot question. The audience walks out unsure what actually happened in the story's central event. This is unresolved-ending, not frayed-ending. The fix: close the central question, fray only peripheral questions.
You leave too many loose threads. Three or four unanswered questions become noise rather than productive thinking-territory. The audience experiences as overload. One or two peripheral loose threads is the working ceiling.
Your fray feels accidental rather than engineered. The audience can't tell whether you chose to leave the thread open or forgot to close it. Make the fray feel deliberate — sometimes through what you almost-say (and as for Jamie...) and then pivot away.
Your ending knots completely. Every question answered, every consequence specified, every character placed. The piece feels closed but doesn't stick. Add a peripheral loose thread.
You fray a piece that needs to knot. Some pieces — particularly highly-comedic or punchline-driven pieces — need clean closure to land. Forcing fray on these pieces produces unsatisfying endings. Tonal calibration matters; not every piece benefits from fray.
The frayed-ending principle is named explicitly in the Knowledge Project interview with the rope-knot metaphor.6 The Thelma-Louise and Butch-Cassidy examples confirm the principle operates at film scale, not just oral-storytelling scale.
A real tension exists with the surprise engine: surprises generally need to close to land. A surprise that frays feels unresolved rather than landed. The principle therefore applies to what comes after the surprise, not to the surprise itself. The surprise-moment closes; the surrounding tissue can fray.
A tension also exists with comedy and joke-form pieces: punchlines need to close completely. The fray principle does not apply to comedy-driven pieces with the same force. Some pieces are better served by full closure than by fray.
Open question: do frayed endings work in short forms (3-minute stories, 30-second social-media pieces)? The mechanism (audience continues thinking after the piece ends) may require sufficient runtime for the audience to have invested enough thinking-attention to want to continue. Ultra-short pieces may not produce the carryover.
Dicks and Karlsson converge cleanly. Karlsson's lowered promises as essay freedom principle treats not delivering on every possible promise as productive — the essay reader's continued thinking is part of the artifact. Both writers recognize that the artifact extends beyond its physical end-point into the audience's continued cognition. The technique-form differs (Karlsson works in essay; Dicks in oral), but the underlying insight is shared.
Dicks and Gelb converge on documentary endings. Gelb's documentary practice often leaves specific small questions unresolved even when the documentary's central question is fully answered. Same principle, different medium.
The frayed-ending is fundamentally an engineered-incompleteness technique — the deliberate leaving of small unresolved threads to extend audience cognition beyond the piece's runtime. That move has direct analogues across disciplines.
Eastern spirituality: Dhvani — Suggestion as Soul of Content — Sanskrit aesthetic theory treats vyanjana (suggestion) as superior to vacya (statement) precisely because suggestion-mode leaves productive incompleteness that the audience continues to work on. The frayed ending is dhvani at the structural level — the piece suggests rather than states its full conclusion, and the audience's continued cognition completes what the piece left open. What neither domain produces alone: classical aesthetics supplies the philosophical depth (why audience-completed art is richer than fully-stated art); Dicks supplies the practitioner technique (leave specific small questions open at the end). The cross-tradition convergence reveals that engineered-incompleteness is at least a thousand years old as an aesthetic discipline, and contemporary working storytelling is rediscovering its operational shape. Working storytellers can borrow the classical articulation to understand why the technique works at a deep level rather than just trusting it as practitioner-folklore.
Behavioral mechanics: Six-Minute X-Ray (Hughes) — Hughes's elicitation work uses unresolved-disclosure moves: the elicitor closes a conversation with a small open question that the target continues thinking about afterward, often producing follow-up disclosures hours or days later when the target reaches out unprompted to fill the loop. The mechanism is structurally parallel to frayed endings: leave specific small loose threads, and the receiver's cognitive completion-seeking extends the engagement beyond the formal end of the interaction. What neither domain produces alone: Dicks teaches the storyteller to use fray for audience-carryover and re-telling; Hughes teaches the elicitor to use fray for delayed-disclosure. Both crafts depend on the receiver's incomplete-pattern-cognition extending beyond the practitioner's last move. Cross-orientation transferability is high — the same technique serves different intents.
The Sharpest Implication Frayed endings imply that the artifact is not the same as the piece. Most amateurs treat a story as the spoken or written content — what happens between the first word and the last word. The principle reveals a different conception: the piece extends into the audience's cognition after delivery ends, and the storyteller can engineer how long the piece keeps running by controlling what threads they leave loose. This re-conceives the boundaries of the work. The piece doesn't end when delivery ends — it ends when audiences stop thinking about it. Skilled storytellers engineer for the longest possible continued-thinking period; amateurs cut off the work at delivery and lose the carryover entirely.
Generative Questions